Overproduction of theBradyrhizobium japonicum c-Type Cytochrome Subunits of thecbb3Oxidase inEscherichia coli

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS ARTICLE NO.

251, 744 –747 (1998)

RC989549

Overproduction of the Bradyrhizobium japonicum c-Type Cytochrome Subunits of the cbb3 Oxidase in Escherichia coli Engin Arslan, Henk Schulz, Rachel Zufferey, Peter Ku¨nzler, and Linda Tho¨ny-Meyer1 Mikrobiologisches Institut, Eidgeno¨ssische Technische Hochschule, Schmelzbergstrasse 7, CH-8092 Zu¨rich, Switzerland

Received September 15, 1998

We report on a system to improve expression of mature c-type cytochromes in Escherichia coli. It is based on the use of plasmid pEC86 that expresses the E. coli cytochrome c maturation genes ccmABCDEFGH constitutively, whereby the production of both endogenous and foreign c-type cytochromes was increased substantially. The periplasmic soluble domains of the c-type cytochrome subunits FixO and FixP of the Bradyrhizobium japonicum cbb3 oxidase could be expressed in E. coli only when pEC86 was provided in a degP-deficient strain. This shows that a stimulation of heme attachment by the Ccm maturase system combined with the diminished proteolytic activity in the periplasm can increase c-type cytochrome yields. © 1998 Academic Press

Escherichia coli produces c-type cytochromes under anaerobic conditions when grown in the presence of nitrite, nitrate or TMAO as the terminal electron acceptors. The covalent attachment of heme to the apocytochromes occurs in the periplasm and requires the function of eight cytochrome c maturation genes, ccmABCDEFGH, which are expressed in the so-called aeg46.5 operon under anaerobic conditions (1,2). These genes encode membrane proteins that comprise an ABC transporter (3–5), a periplasmically oriented thioredoxin system (6), a heme chaperone (7), and a putative heme lyase (3–5). It was shown previously that a chromosomal deletion of the ccm genes led to a complete loss of holocytochrome c formation (1,2). When the ccm genes were provided on a plasmid together with the structural genes for the NapB and NapC c-type cytochromes of the periplasmic nitrate reduc1

To whom correspondence should be addressed. Fax: (41) (1) 632 11 48. E-mail: [email protected]. Abbreviations used: PCR, polymerase chain reaction; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; TCA, trichloroacetic acid; TMAO, trimethylamine-N-oxide. 0006-291X/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

tase, cytochrome c maturation was restored and even seemed to be increased (8). Cytochromes of the c-type are interesting molecules for structural and spectroscopic studies. Therefore, a system in which they can be overproduced is highly desirable. Overproduction of various heterologous c-type cytochromes has been attempted in E. coli, however, with variable success (9). For example, we were able to produce the soluble holocytochrome c550 of Bradyrhizobium japonicum (10) in E. coli; yet, this was achieved only under anaerobic, nitrate-dependent growth conditions, and the yield was rather low (11). While studying the B. japonicum cbb3 type heme copper oxidase (12) it became of particular interest to overproduce the mono- and diheme cytochrome c subunits II (FixO) and III (FixP) in E. coli in order to analyze their biochemical properties. Here we present an expression system for c-type cytochromes that makes use of a plasmid from which the ccm genes are expressed constitutively and independently of oxygen control. Co-expression of the ccm genes together with a structural gene for apocytochrome c can lead to high levels of holocytochrome c. To produce significant amounts of mature FixO and FixP in E. coli, it was necessary to use a strain in which the gene encoding the periplasmic protease DegP was defective (13). MATERIALS AND METHODS Bacterial strains and growth conditions. E. coli MC1061 (hsdR araD139 D(araABC-leu)7679 galU galK D(lac) FX74 rpsL thi) (14) and HM125 (F2 DlacX74 galE galK thi rpsL(strA) DphoA degP41::VKanr eda51::Tn10(Tetr) rpoH15 (15) were used as hosts for cytochrome c expression. Aerobic growth was in LB medium (16), anaerobic growth in MS medium with either 20 mM NO2 3 , 2.5 mM NO2 2 or 20 mM TMAO (17). Antibiotics were added at the following final concentrations: ampicillin 200 mg ml21, chloramphenicol 10 mg ml21. Cells were grown at 37°C for expression of the B. japonicum cytochrome c550 and the B. subtilis cytochrome c-550.Cells were grown at 30°C for expression of FixO and FixP. Cytochrome c-specific gene expression was induced at a cell density of A600 5 0.6 with 0.8% arabinose for 4 h.

744

Vol. 251, No. 3, 1998

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 1. Overproduction of B. japonicum cytochrome c550 in aerobically (A) or anaerobically (B) grown E. coli. Periplasmic protein obtained by small-scale preparations was TCA precipitated and separated by 15% SDS–PAGE. The c-type cytochromes were visualized by heme staining. Construction of plasmids. For the construction of pEC86, a PstI linker was fused in a first step to the EcoRV site of the pACYC184 vector in pEC66 (11), resulting in pEC83. This plasmid contained the napBCccmABCDEFGH sequence in the tet gene of the vector in the same orientation, thus allowing outreading transcription from the tet promoter. It was opened with PstI/AflII to delete the napBCccmAB’ segment, and the 1.2-kb NsiI/AflII fragment containing ccmAB’ was inserted to restore an intact ccmABCDEFGH sequence. To express a soluble FixO protein (FixOsol), a StuI site was introduced at codons 33–35 of the fixO sequence by PCR and ligated with a StuI site at the end of the coding sequence for the OmpA signal peptide (21 codons). This resulted in an additional serine codon between the OmpA coding sequence and L36 codon of fixO. The ATG start codon of the ompA gene was engineered into a NdeI site by PCR. The ompA’-‘fixO fusion was then cloned as an NdeI-XbaI fragment into the expression vector pISC-2 downstream of the arabinose-inducible para promoter to give pRJ4646 (11). A ompA’-‘fixP fusion was constructed on pRJ4591 using the same strategy. In this case the OmpA signal sequence was fused to FixP54-290 at a StuI site introduced at codon 53 of FixP, again creating an additional serine codon. All PCR products and fusion sites were confirmed by sequencing. Preparation of membranes and periplasmic fraction. Membranes were prepared from 1 l of stationary-phase grown cells as described previously (8). Periplasmic fractions were prepared from 250 ml of aerobically grown cells induced for cytochrome c expression using lysozyme (11). For small-scale preparations, the treatment was applied to cells of 10 ml cultures of an A600 5 0.6, and the entire periplasmic fraction was precipitated with TCA. Characterization of c-type cytochromes. Protein determination, SDS–PAGE, heme stains using 3,39-dimethoxybenzidine, and Western blot analyses with antibodies against FixO and FixP were performed as described (18), except that detection was done with 3-(4-methoxyspiro{1,2-dioxetan-3,29-(59chloro)tricyclo[3.3.1.13,7]decan}-4-yl)phenylphosphate (CSPD; Boehringer, Mannheim, Germany) and goat antirabbit IgG alkaline phosphatase conjugate (BioRad, Glattbrugg, Switzerland) as the secondary label. Optical difference spectra were recorded on a Hitachi model U-3300 spectrophotometer (12).

ground under anaerobic growth conditions in the presence of nitrate. Under aerobic growth conditions, no holocytochrome c550 was obtained. We reasoned that under such growth conditions a limitation of ccm gene products, which are expressed from the Fnr-dependent aeg promoter (19), may have caused the absence of cytochrome c maturation. Therefore, we constructed a plasmid that expressed the ccm genes from a Fnrindependent promoter. Plasmid pEC86 is derived from the vector pACYC184 and contains the ccm genes downstream of the tet promoter. Unlike the previously described plasmid pEC66 (8), it did not contain the napBC genes encoding two of the endogenous E. coli c-type cytochromes and was therefore more suitable to enhance the formation of heterologous c-type cytochromes. Fig. 1A shows that pEC86 can in fact induce maturation of the B. japonicum cytochrome c550 even under aerobic conditions. Stimulation of cytochrome c maturation was also obtained under anaerobic conditions in the presence of nitrate (Fig. 1B). In this case, the chromosomally expressed napB gene product was also produced at detectable levels in the periplasm (Fig. 1B, lane 2). Expression of a membrane-bound c-type cytochrome. It was shown previously that expression of the membrane-bound Bacillus subtilis cytochrome c-550 (encoded by cccA) in E. coli was possible under aerobic conditions (20), but required the presence of the ccm gene cluster (1). Although expression of the nap-ccm operon is induced anaerobically, sufficient amounts of Ccm polypeptides appear to be made under aerobic conditions to support maturation of the CccA polypeptide. We tried to stimulate the production of B. subtilis cytochrome c-550 by co-expressing the ccm genes on pEC86. Fig. 2 shows that similar levels of this cytochrome were produced independently of the presence of pEC86. When the ccm genes were co-expressed, the membranes accumulated the heme-binding form of the heme chaperone CcmE (7), as visualized by heme staining. This result indicates that expression of the cccA gene from its own Bacillus promoter may be the limiting step for overproduction.

RESULTS Overproduction of c-type cytochromes with pEC86 encoding the ccm genes. We have previously established an expression system for the soluble monoheme cytochrome c550 of B. japonicum in E. coli (11). The structural gene cycA was expressed from an arabinoseinducible promoter para in an E. coli MC1061 back-

FIG. 2. Overproduction of B. subtilis cytochrome c-550 in E. coli. Membrane proteins (40 mg per lane) were separated by 15% SDS– PAGE and proteins with covalently bound heme were visualized by heme staining.

745

Vol. 251, No. 3, 1998

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 3. Overproduction of soluble versions of the B. japonicum FixO and FixP proteins. Top panel: heme stain of periplasmic fractions (20 mg per lane) with overproduced FixOsol from pRJ4646 and FixPsol from pRJ4591, respectively. Purified cbb3-type cytochrome oxidase (2 mg) was loaded on the right as control (lane 5). Below are Western blots of identical gels probed with anti-FixO (middle panel) and anti-FixP (bottom panel) immunoglobulins (bottom panel).

Overproduction of the subunits FixO and FixP of the B. japonicum cbb3-type oxidase. We were particularly interested to overproduce the B. japonicum c-type cytochrome subunits of the cbb3 oxidase as individual proteins. To facilitate overproduction of these cytochromes that are normally membrane-anchored by an N-terminal, hydrophobic transmembrane helix, we constructed soluble, periplasmic versions by replacing the transmembrane helices with a cleavable OmpA signal sequence. It was shown previously, that FixO and FixP are extremely unstable in B. japonicum if their assembly with cofactors and other subunits of the FixNOP complex was disturbed (18). A soluble version of FixO was constructed by fusing the cleavable OmpA signal sequence to Leu36 of FixO. Expression of the fused gene was again from the para promoter. No mature FixO protein was obtained in the periplasmic fraction of induced, anaerobically grown MC1061 cells. However, when whole cell extracts of induced cells were loaded directly on an SDS polyacrylamide gel, a weak signal was obtained in a Western blot (not shown), which indicated that apocytochrome was formed and rapidly degraded. To improve holocytochrome c formation, fixO was co-expressed with the ccm genes on pEC86; however, maturation of FixOsol was still not observed (not shown). Next we combined this approach with the use of a strain that was deficient in the periplasmic protease DegP (14). Mature FixO protein was detected as a 25-kDa heme-staining protein in periplasmic fractions of induced cells (Fig. 3, top panel, lane 2), which was in agreement with the theoretical molecular mass of 23’548 Da of FixOsol protein. Maximal expression was obtained after 4h of induction with 0.8% arabinose (not shown). Western blot analysis (Fig 3, middle panel, lane 2) confirmed the identity of overexpressed FixOsol.; As sometimes observed with cytochromes c, the protein migrated as a double band. Can this system also be used for overproduction of the diheme cytochrome FixP? A similar construct with

the cleavable OmpA signal sequence replacing the N-terminal membrane anchor amino acids (1–53) of Fix P was introduced into the degP¸ strain, and cytochrome c expression was induced with arabinose. Again, FixP holoprotein was detected in the periplasm. The protein detected by heme stain (Fig. 3, top panel, lane 4) and Western blot analysis (Fig. 3, bottom panel, lane 4) migrated as a double band at about 25 kDa, again corresponding to the predicted molecular mass of 24’868 Da of the FixPsol polypeptide. The soluble FixO and FixP proteins were characterized spectrophotometrically. Reduced minus oxidized difference spectra of periplasmic fraction containing either FixO or FixP are shown in Fig. 4. The a-absorption maxima of the difference spectrum of both proteins were at 551 nm, which is in good agreement with the symmetric peak at the same wavelength of the purified cbb3-type cytochrome oxidase that contains all three hemes C (12). We obtained only for the FixPsol, but not for the FixOsol protein, a typical CO-difference spectrum that has also been described for the entire cbb3-type oxidase. We conclude that the FixP protein is the CO-reactive c-type cytochrome of this oxidase. The yield of FixOsol and FixPsol was calculated to be 0.34 mg and 0.18mg, respectively, per ml periplasmic fraction, using a molar extinction coefficient of 19.1 mM21 cm21 (20). These values corresponded to 3.5% and 2.4%, respectively, of the total periplasmic protein. DISCUSSION We have established a system that allows overproduction and maturation of various c-type cytochromes in E.

FIG. 4. Spectrophotometric characterization of FixOsol and FixPsol. A and B are the dithionite-reduced minus air-oxidized absorption difference spectra of periplasmic fractions with overproduced FixOsol (0.49 mg ml21 periplasmic protein) and FixPsol (0.38 mg ml-1 periplasmic protein), respectively. C, dithionite 1 CO minus dithionite-reduced difference spectrum of the same fraction as in B.

746

Vol. 251, No. 3, 1998

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

coli. It is based on two characteristics known to play a role in cytochrome c biogenesis. First, some apocytochromes appear to be extremely unstable polypeptides that are rapidly degraded in the periplasm if they do not bind heme. The use of a degP2 strain reduces periplasmic proteolytic activity and thus may enhance the chances of heme binding. Second, high amounts of ccm gene products have been observed to stimulate heme attachment. The recently identified function of CcmE as a periplasmic heme chaperone that binds heme temporarily and releases it when apocytochrome c substrate is available (7) illustrates that restricted ccm gene expression might be limiting for the production of large amounts of holocytochrome c. Plasmid pEC86 provides a tool for constitutive ccm gene expression and in particular facilitates aerobic cytochrome c maturation. It can also be used to increase the amounts of endogenous c-type cytochromes. In this work, we have used the degP2/pEC86 system to produce mature FixO and FixP proteins. These proteins are the subunits of the B. japonicum cbb3-type oxidase and are thought to be the electron donors for the reduction of O2 at the binuclear heme-copper center in FixN. It has not been possible to obtain mature FixO or FixP protein in the absence of the FixN subunit, which by itself is a b-type cytochrome. The reason for this may be lack of stability of the isolated subunit (18). When these subunits were overexpressed individually in E. coli we observed again a great instability of the apocytochromes. Mature holocytochromes could not be detected unless a degP2 mutant strain was used to protect the polypeptides from periplasmic proteolysis. We succeeded at producing 1.36 mg FixO and 0.37 mg FixP per liter of cultured cells, which was sufficient to determine the spectroscopic characteristics of these proteins. In the reduced minus oxidized difference spectrum they showed a-bands with a maximum at 551 nm, respectively. FixP was CO-reactive, thus confirming the through at 552 nm that had been detected previously for the entire cbb3-type oxidase (12). The B. subtilis membrane-bound cytochrome c-550 is expressed and matured in aerobically grown E. coli without pEC86. Although maturation of this cytochrome c depends on the ccm genes (1), no stimulation of cytochrome c-550 production in the presence of pEC86 was observed. This indicates that low levels of ccm gene products must be synthesized already under aerobic conditions, and these are sufficient for maturation of the entire pool of B. subtilis CccA polypeptide. The CccA polypeptide appears to be relatively stable even when heme is not bound to it, perhaps due to its localization to the membrane. By contrast, the B. japonicum CycA apocytochrome was shown to be extremely sensitive towards periplasmic degradation (11). The latter cytochrome may be protected from proteolysis by interaction with, and heme incorporation by, the Ccm machinery, whereas the former may use the Ccm-maturase system only for heme attachment and not for additional stabilization.

In summary, plasmid pEC86 has been proven to be a valuable tool to enhance maturation and, therefore, production of soluble cytochromes of the c-type produced in E. coli. This will be of particular interest for applications such as structural and spectroscopic characterizations that require large amounts of mature protein to be produced from minimal culture volumes. ACKNOWLEDGMENTS We thank G. Georgiou for providing the degP2 strain and A. Hungerbu¨hler for excellent technical assistance. H. Hennecke is greatly acknowledged for advice and generous support. This work was supported by the Swiss National Foundation for Scientific Research.

REFERENCES 1. Tho¨ny-Meyer, L., Fischer, F., Ku¨nzler, P., Ritz, D., and Hennecke, H. (1995) J.Bacteriol. 177, 4321– 4326. 2. Grove, J., Tanapongpipat, S., Thomas, G., Griffiths, L., Crooke, H., and Cole, J. (1996) Mol. Microbiol. 19, 467– 481. 3. Tho¨ny-Meyer, L. (1997) Microbiol. Mol. Biol. Rev. 61, 337–376. 4. Page, M. D., Sambongi, Y., and Ferguson, S. J. (1998) Trends Biochem. Sci. 23, 103–108. 5. Kranz, R. G., Lill, R., Goldman, B., Bonnard, G., and Merchant, S. (1998) Mol. Microbiol. 29, 383–396. 6. Fabianek, R., Hennecke, H., and Tho¨ny-Meyer, L. (1998) J. Bacteriol. 180, 1947–1950. 7. Schulz, H., Hennecke, H., and Tho¨ny-Meyer, L. (1998) Science 281, 1197–1200. 8. Tho¨ny-Meyer, L., and Ku¨nzler, P. (1997) Eur. J. Biochem. 246, 794 –799. 9. Tho¨ny-Meyer, L., Ritz, D., and Hennecke, H. (1994) Mol. Microbiol. 12, 1–9. 10. Bott, M., Tho¨ny-Meyer, L., Loferer, H., Rossbach, S., Tully, R. E., Keister, D., Appleby, C. A., and Hennecke, H. (1995) J. Bacteriol. 177, 2214 –2217. 11. Tho¨ny-Meyer, L., Ku¨nzler, P., and Hennecke, H. (1996) Eur. J. Biochem. 235, 754 –761. 12. Preisig, O., Zufferey, R., Tho¨ny- Meyer, L., Appleby, C. A., and Hennecke, H. (1996) J. Bacteriol. 178, 1532–1538. 13. Strauch, K. L., Johnson, K., and Beckwith, J. (1989) J. Bacteriol. 171, 2689 –2696. 14. Meissner, P. S., Sisk, W. P., and Bergman, M. L. (1987) Proc. Natl. Acad. Sci. USA 84, 4171– 4175. 15. Hendrikus, J. M., and Georgiou, G. (1994) Bio/Technology 12, 1107–1110. 16. Miller, J. H. (1992) A Laboratory Manual and Handbook for Escherichia coli and Related Bacteria, Cold Spring Harbour Laboratory Press, Cold Spring Harbor, NY. 17. Iobbi-Nivol, C., Crooke, H., Griffith, L., Grove, J., Hussain, H., Pommier, J., Mejean, V., and Cole, J. A. (1994) FEMS Microbiol. Lett. 119, 89 –94. 18. Zufferey, R., Preisig, O., Hennecke, H., and Tho¨ny-Meyer, L. (1996) J. Biol. Chem. 271, 9114 –9119. 19. Choe, M., and Reznikoff, W. S. (1993) J. Bacteriol. 175, 1165– 1172. 20. Jones, C. W., and Poole, R. K. (1985) Methods Microbiol. 18, 285–328. 21. Von Wachenfeldt, C., and Hederstedt, L. (1990) FEBS Lett. 270, 147–151.

747